Health Affairs, 25, no. 2 (2006): 437-443 doi: 10.1377/hlthaff.25.2.437 © 2006 by Project HOPE	New Online   McAllen, TX & Beyond: An Expert Roundtable   Geography & Reform   Medicaid or Insurance Exchange?   Siren Song of New GME   Public Plan Option: Pro & Con

This Article Abstract Reprint (PDF) Technical Appendix Submit a response to this article Comments: View responses Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (7) Citing Articles via Google Scholar Google Scholar Articles by Jacobson, M. Articles by Newhouse, J. P. Search for Related Content PubMed PubMed Citation Articles by Jacobson, M. Articles by Newhouse, J. P. Related Collections Legal/Regulatory Issues Medicare Business Of Health Pharmaceuticals Quality Of Care Research And Technology Health Spending Consumer Issues

Health Tracking

MarketWatch

Does Reimbursement Influence Chemotherapy Treatment For Cancer Patients?

Abstract

 
Before the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003, Medicare reimbursed physicians for chemotherapy drugs at rates that greatly exceeded physicians’ costs for those drugs. We examined the effect of physician reimbursement on chemotherapy treatment of Medicare beneficiaries older than age sixty-five with metastatic lung, breast, colorectal, or other gastrointestinal cancers between 1995 and 1998 (9,357 patients). A physician’s decision to administer chemotherapy to metastatic cancer patients was not measurably affected by higher reimbursement. Providers who were more generously reimbursed, however, prescribed more-costly chemotherapy regimens to metastatic breast, colorectal, and lung cancer patients.

Although the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) altered the structure of chemotherapy reimbursements, so that physicians are now paid based on manufacturers’ average sales price (ASP) plus 6 percent and an administrative fee, data exploring the relationship between various reimbursement incentives and practice patterns are still important. First, they provide clues for how MMA might change chemotherapy treatment for Medicare beneficiaries. Second, since reimbursement rates for chemotherapy vary greatly within the private market, with many insurance companies basing their rates on the AWP, this research could uncover treatment distortions that persist in the private market. More generally, this work should help us understand the extent to which financial incentives—in this case, unintentional ones—can affect physicians’ clinical decisions.

In this paper we analyze the relationship between physicians’ prescribing decisions and Medicare reimbursement. An alternative would have been to analyze the spread between Medicare reimbursement and the prices at which oncologists purchased the drugs, but we could not observe purchase prices. Nonetheless, because chemotherapy drugs can be bought directly from manufacturers or through national group purchasing organizations, similar types of clinics or physicians in terms of bargaining power and volume of business should have been able to purchase drugs at similar prices.⁴ Thus, variation in reimbursement—the variable we analyze—should track closely with variation in physicians’ profit from dispensing drugs.

Specifically, we asked two questions: Were physicians who were more highly reimbursed or who experienced greater increases in their reimbursement more likely to prescribe chemotherapy? And, conditional on prescribing chemotherapy, were such physicians more likely to use expensive drugs?

Study Methods And Data

Top Study Methods And Data Results Discussion NOTES

 
Study methods. To assess how reimbursement affected treatment, we exploited differences in Medicare reimbursement rates across physicians at a point in time and for the same physician over time. Variation at a point in time stemmed from local carriers’ discretion before October 2002 to determine the composition of the Healthcare Common Procedure Coding System (HCPCS), the basis for Medicare reimbursement. In particular, individual carriers processing Part B claims chose the specific National Drug Code (NDC, corresponding to a unique chemical entity, form, strength, package size, manufacturer, and AWP) that determined the reimbursement rate for all NDCs in the HCPCS. Because their choices varied, the reimbursement for a HCPCS covering a multisource drug, or a single-source drug available in different forms or package sizes, varied across carriers.⁵ Although such variation was typically on the order of 10 percent or less, it was much larger for some commonly prescribed chemotherapy drugs.⁶ For example, in 1999 the spread between the highest and lowest reimbursement for 10 mg doxorubicin was about 27 percent ($52.44 versus $38.03). Moreover, these reimbursements changed differentially over time based on revisions to the AWP for specific NDCs.

Data sources. Study cohort. To study chemotherapy treatment, we analyzed the Surveillance, Epidemiology, and End Results (SEER) cancer registries and linked Medicare claims.⁷ Our sample consisted of all Medicare-eligible patients older than age sixty-five with lung, breast, colorectal, or other gastrointestinal (GI—pancreatic, esophageal, liver, and stomach) cancers who had metastatic cancer between 1995 and 1998 and had filed Medicare claims during this period.⁸

We chose these cancers because they account for more than 60 percent of cancer deaths in North America and, when they are metastatic, chemotherapy is a primary component of their treatment.⁹ We did not use data from before 1995 because of changes in recommended treatment. We defined metastasis as presentation with Stage IV disease, as documented by SEER abstractors within four months of diagnosis, or documentation of two or more Medicare claims separated by thirty days for a secondary cancer site.¹⁰

We studied only patients with metastatic disease because such patients have a relatively short expected survival time, regardless of treatment. In other words, we expected little effect on outcomes from any variation in treatment. Moreover, treatment options are less standardized than in early stages, when cure is the goal. Thus, observed treatment differences across metastatic patients, in contrast to those at earlier stages of the disease, are more likely to result from physicians’ decisions and patients’ preferences than from unobservable differences in health. And since patients’ preferences are unlikely to vary systematically with financial incentives, any relationship between treatment and reimbursement is most likely attributable to the physician’s response.

Outcomes studied. We analyzed whether the patient received chemotherapy treatment in the three-month period after a metastatic cancer diagnosis, conditional on the patient’s surviving at least twenty-eight days, not having been in chemotherapy treatment three months prior to this diagnosis, and not entering hospice. Patients were deemed to have received any chemotherapy if they had a single claim for a chemotherapy-related code.¹¹

Specific agents (J9s) were identified from outpatient settings (Physician/Supplier [NCH] or Outpatient [OUTPT] files); drugs administered to inpatients are not recorded on the claim because they are bundled with the diagnosis-related group (DRG) payment. We included anti-emetics that could be administered intravenously and were thus reimbursed by Medicare, as well as calcium leucovorin, which is given intravenously with chemotherapy agents such as fluorouracil.¹²

In addition to analyzing whether reimbursement affected the likelihood of chemotherapy, we also analyzed the costliness of the agents used when chemotherapy was prescribed. Reimbursement amounts were garnered from the NCH files only, because drug payments cannot be identified in institutional claims. We excluded the roughly 15 percent of reimbursements for (J9) chemotherapy claims with unidentified agents (J9999) because we could not determine the drugs used or their standard dosages.

To remove variability in spending caused by patient-specific dosing—which is affected by the patient’s height, weight, and organ functioning—and the number of doses a patient received—which is affected by therapy tolerance and response—we defined the initial treatment regimen as the combination of chemo-therapeutic agents, anti-emetics, and leucovorin administered to the patient within the first twenty-eight days after a metastatic cancer diagnosis. Each regimen was treated as if a standard quantity were administered by assuming that (1) each drug was given according to the most common dose and schedule recommended in the Dana-Farber Cancer Institute’s Oncology Protocol System; and (2) each patient had a body surface area of 1.7 m². Anti-emetics were dosed in combination with the highest-frequency-dosed chemotherapy agent a patient received.

Each standardized dose was then priced using the average "national" (across all SEER regions) reimbursement for that agent within a tumor site and year and summed for all drugs a patient received.¹³ Because we used national average prices to determine expenditure, any variation across patients is attributable to the drugs prescribed rather than the prices paid to physicians. Furthermore, by standardizing dosages, this measure captures a physician’s initial regimen choice, rather than any adjustments dictated by a patient’s tolerance of and response to treatment.

Explanatory variables. To measure how generously each physician was reimbursed, we defined a summary measure or index for the regimens prescribed for a given tumor site in a given year. The index was the sum of the weighted average difference between the physician’s and the national mean reimbursement for each agent the physician prescribed, where the weights were the ratio of national spending on a regimen to total spending on all chemotherapy regimens for that tumor site.¹⁴ In some analyses we deflated the index by Medicare’s 1997 Geographic Practice Cost Index (GPCI) for the relevant region, to account for variation across regions in the cost of administration. In summary, this index captured how generously an individual physician was reimbursed relative to the national average for the set of agents that physician prescribed.

Statistical analyses. To model the impact of reimbursement generosity on a patient’s likelihood of chemotherapy treatment, we used probit regressions with provider-specific random effects. Our models include year-indicator variables to control for national trends in treatment patterns.¹⁵ We also controlled for patients’ age and age squared, year of meta-static cancer diagnosis, sex, race (white, black, Asian, Hispanic, or other), marital status (single, married, separated/divorced, or widowed), tumor grade (well, moderately, or poorly differentiated; undifferentiated; or unknown), census tract per capita income or the state average if missing; and a dummy variable for whether the state average income was used. Comorbidities were identified from diagnostic billing codes in both inpatient and outpatient claims, assigned scores based on severity, and then summed to form an index, called the Deyo-Charlson score.¹⁶ When we estimated results across all tumor sites, we also included indicator variables for tumor site.

To estimate the impact of the reimbursement generosity on the costliness of chemo-therapeutic agents prescribed, we used linear regressions with provider-specific random effects. We conducted our analysis pooling across tumor sites as well as separately by site, since the types and costs of regimens varied markedly by cancer site. We included year-indicator variables in all models and controlled for the same set of patient characteristics as in the probability-of-chemotherapy models.

Results

Top Study Methods And Data Results Discussion NOTES

 
Chemotherapy use. Almost half of the sample presented with Stage IV disease at diagnosis (SEER stage distant), although this proportion varied considerably across tumor sites. About 40 percent of the sample was treated with chemotherapy (Exhibit 1). Breast cancer patients were much less likely than others to receive chemotherapy, possibly because of the hormone therapy option, which we cannot observe in Medicare claims. Lung cancer patients also had relatively low chemotherapy rates, likely because chemotherapy was just gaining acceptance for this group during the study period.¹⁷

Costliness of agents used. Monthly chemotherapy spending varied markedly across tumor sites (Exhibit 1). Patients with metastatic breast and colorectal cancer, on average, received regimens that cost less than $1,000 per month, whereas the regimens prescribed to other GI cancer patients cost on average $1,400 per month and to lung cancer patients, more than $3,600 per month.

The reimbursement index had a clear effect on the costliness of chemotherapeutic agents prescribed (Exhibit 3). More generously reimbursed providers prescribed more-costly regimens to breast (p < .038), colorectal (p < .079), and lung (p < .039) cancer patients. In contrast, more generously reimbursed providers prescribed less costly regimens to metastatic GI cancer patients (p < .038). For breast cancer patients, a one-dollar increase in a physician’s reimbursement resulted in the use of agents that cost twenty-three dollars more. Another way to interpret the values in the exhibit is to estimate the effect of a one-standard-deviation increase in the index. Among breast cancer patients receiving chemotherapy, a one-standard-deviation increase in the reimbursement index (2.89, data not shown) was associated with a $67 increase in spending on chemotherapeutic agents (evaluated by multiplying 2.89 by the regression coefficient of 23.1). For colon and lung cancer patients, a one-standard-deviation increase in physicians’ reimbursement index (1.14 and 33.9, respectively) raised spending by $40 and $150, respectively. With average chemotherapy spending of $858, $705, and $3,772 for breast, colon, and lung cancer patients, respectively, such increases would raise spending 4–8 percent.

Discussion

Top Study Methods And Data Results Discussion NOTES

 
We found no evidence that reimbursement incentives affected oncologists’ decisions to administer chemotherapy to metastatic cancer patients. Once a decision to give chemotherapy was taken, however, physicians receiving more-generous Medicare reimbursements used more-costly treatment regimens. Except for the noncolorectal GI cancer site, which was so heterogeneous that such patients might not have been well characterized in our data, these results were similar whether aggregated or stratified by tumor site.

Study limitations. Our study has some limitations. To ascertain chemotherapy, we relied on claims data, which are not created for research and could be less accurate than desired. Claims-based comorbidity measurement is not the same as performance status, on which clinicians make treatment decisions.

The use of Medicare data limited us to elderly patients, and practice patterns and incentives might differ for younger, commercially insured patients. Nonetheless, many commercial insurance companies also base reimbursement rates on AWP. A survey of thirty-two health plans found that many reimbursed for chemotherapy drugs at 95 percent of AWP, but others reimbursed at rates as low as 75 percent, and still others at rates as high as 125 percent.¹⁸ If our results generalize to the commercially insured, they suggest that the administration of chemotherapy to such patients should be little affected by such variation, but the mix of agents may well respond.

Policy implications. As noted in the introduction, Medicare no longer uses AWP as a basis for reimbursement. Our results suggest that rates of chemotherapy administration will not change much, provided that oncologists continue to accept Medicare patients. On the other hand, since physicians will no longer differentially profit from using particular agents, this new reimbursement method could modify the mix of chemotherapy drugs used.

Oncologists are loath to acknowledge that financial motives can affect treatment decisions. Although reimbursement seems to have little effect on the primary decision to administer palliative chemotherapy to patients with advanced solid tumors, it appears to affect the choice of drugs used.

Oncologists have maintained that in the past, the markup on the drugs compensated for Medicare’s prior failure to reimburse for the cost of administering the drug. In response, Medicare now pays a fee for administration but, as noted above, reimburses the drug at a 6 percent markup over what the physician paid for it. These changes should make choice of agents based more on clinical considerations and patients’ preferences and less on reimbursement decisions.

Editor's Notes

 
Mireille Jacobson (mgjacobs{at}umich.edu) is a Robert Wood Johnson Foundation Research Fellow at the University of Michigan in Ann Arbor. She is on leave from the Department of Planning, Policy, and Design at the University of California, Irvine, where she is an assistant professor. James O’Malley is an assistant professor of statistics in the Department of Health Care Policy, Harvard Medical School, in Boston, Massachusetts. Craig Earle is an associate professor of medicine at the medical school and an oncologist at the Dana-Farber Cancer Institute in Boston. Juliana Pakes is a data analyst and project administrator in the Department of Health Care Policy, Harvard Medical School. Peter Gaccione is a programmer who worked as a consultant in the department. Joseph Newhouse is the John D. MacArthur Professor of Health Policy and Management at Harvard University.

Earlier versions of this study were presented at the 2003 Annual American Economics Association Meetings, the 2003 National Bureau of Economic Research (NBER) Health Care Program Meeting, and the Harvard Medical School Department of Health Care Policy P01 Seminar. The authors thank the meeting participants and the anonymous referees for their many helpful comments. They also gratefully acknowledge research support from the Agency for Healthcare Research and Quality (Grant no. P01-HS-10803).

NOTES

Top Study Methods And Data Results Discussion NOTES

 

1. U.S. Department of Health and Human Services, Excessive Medicare Payment for Prescription Drugs, Pub. no. OEI-03-97-00290 (Washington: U.S. Government Printing Office, December 1997); and DHHS, Medicare Reimbursement of Prescription Drugs, Pub. no. OEI-03-00-00310 (Washington: GPO, December 2001).
2. DHHS, "HCFA Program Memorandum," AB-99- 63 (Washington: GPO, September 1999); and Medical Economics Company, Drug Topics Red Book (Montvale, N.J.: Medical Economics Co., various years).
3. U.S. Government Accountability Office, Medicare: Payments for Covered Outpatient Drugs Exceed Provider’s Cost, Pub. no. GAO-01-1118 (Washington: GPO, September 2001).
4. DHHS, Medicare Reimbursement, 2001.
5. Ibid.
6. DHHS, "HCFA Program Memorandum."
7. L.A. Ries et al., SEER Cancer Statistics Review, 1973–1994, Pub. no. 97-2789 (Bethesda, Md.: National Cancer Institute, 1997); A.B. Nattinger, T.L. McAuliffe, and M.M. Schapira, "Generalizability of the Surveillance, Epidemiology, and End Results Registry Population: Factors Relevant to Epidemiologic and Health Care Research," Journal of Clinical Epidemiology 50, no. 8 (1997): 939–945[CrossRef][Web of Science][Medline]; and A.L. Potosky et al., "Potential for Cancer Related Health Services Research Using a Linked Medicare-Tumor Registry Database," Medical Care 31, no. 8 (1993): 732–748. Also, see the online technical appendix at http://content.healthaffairs.org/cgi/content/full/25/2/437/DC1.[Web of Science][Medline]
8. We excluded people age sixty-five and younger because their course of chemotherapy treatment might have been determined prior to their Medicare eligibility.
9. R.T. Greenlee et al., "Cancer Statistics, 2000," Cancer Journal for Clinicians 50, no. 1 (2000): 7–33.
10. C.C. Earle et al., "Identifying Cancer Relapse Using SEER-Medicare Data," Medical Care 40, no. 8 Supp. (2002): IV-75–IV-81.
11. J.L. Warren et al., "Utility of the SEER-Medicare Data to Identify Chemotherapy Use," Medical Care 40, no. 8 Supp. (2002): IV-55–IV-61; U.S. Public Health Service, International Classification of Diseases, Ninth Revision, Clinical Modification (ed. 5) (Los Angeles: Practice Management Information Corp., 1996); American Medical Association, Physician’s Current Procedural Terminology: CPT 94 (Chicago: AMA, 1993); Centers for Medicare and Medicaid Services, HCFA Common Procedure Coding System (HCPCS): National Level II Medicare Codes (Los Angeles: Practice Management Information Corp., 1994); and CMS, HCFA Data Dictionary: Revenue Center Codes (Baltimore: CMS, 17 June 1999). See also the online appendix, as in Note 7.
12. Advanced Colorectal Cancer Meta-Analysis Project, "Modulation of Fluorouracil by Leucovorin in Patients with Advanced Colorectal Cancer: Evidence in Terms of Response Rate," Journal of Clinical Oncology 10, no. 6 (1992): 896–903[Abstract]; and C.H. Kohne et al., "Effective Biomodulation by Leucovorin of High-Dose Infusion Fluorouracil Given as a Weekly Twenty-Four-Hour Infusion: Results of a Randomized Trial in Patients with Advanced Colorectal Cancer," Journal of Clinical Oncology 16, no. 2 (1998): 418–426.[Abstract]
13. See the online technical appendix, as in Note 7.
14. Ibid.
15. Ibid.
16. R.A. Deyo, D.C. Cherkin, and M.A. Ciol, "Adapting a Clinical Comorbidity Index for Use with ICD-9-CM Administrative Databases," Journal of Clinical Epidemiology 45, no. 6 (1992): 613–619[CrossRef][Web of Science][Medline]; and M.E. Charlson et al., "A New Method of Classifying Prognostic Comorbidity in Longitudinal Studies: Development and Validation," Journal of Chronic Diseases 40, no. 5 (1987): 373–383.[CrossRef][Web of Science][Medline]
17. Non–Small Cell Lung Cancer Collaborative Group, "Chemotherapy in Non–Small Cell Lung Cancer: A Meta-Analysis Using Updated Data on Individual Patients from Fifty-two Randomised Clinical Trials," British Medical Journal 311, no. 7010 (1995): 899–909.[Abstract/Free Full Text]
18. Medicare Payment Advisory Commission, "Medicare Payments for Outpatient Drugs," in Report to the Congress: Variation and Innovation in Medicare (Washington: MedPAC, June 2003), chap. 9, Part B.

                      What's this?

This article has been cited by other articles:

				P. B. Bach Limits on Medicare's Ability to Control Rising Spending on Cancer Drugs N. Engl. J. Med., February 5, 2009; 360(6): 626 - 633. [Full Text] [PDF]

				J. McFarlane, J. Riggins, and T. J. Smith SPIKE$: A Six-Step Protocol for Delivering Bad News About the Cost of Medical Care J. Clin. Oncol., September 1, 2008; 26(25): 4200 - 4204. [Full Text] [PDF]

				C. C. Earle, M. B. Landrum, J. M. Souza, B. A. Neville, J. C. Weeks, and J. Z. Ayanian Aggressiveness of Cancer Care Near the End of Life: Is It a Quality-of-Care Issue? J. Clin. Oncol., August 10, 2008; 26(23): 3860 - 3866. [Abstract] [Full Text] [PDF]

				S. E. Harrington and T. J. Smith The Role of Chemotherapy at the End of Life: "When Is Enough, Enough?" JAMA, June 11, 2008; 299(22): 2667 - 2678. [Abstract] [Full Text] [PDF]

				S. D. Ramsey How Should We Pay the Piper When He's Calling the Tune? On the Long-Term Affordability of Cancer Care in the United States J. Clin. Oncol., January 10, 2007; 25(2): 175 - 179. [Full Text] [PDF]

				N. J. Meropol and K. A. Schulman Cost of Cancer Care: Issues and Implications J. Clin. Oncol., January 10, 2007; 25(2): 180 - 186. [Abstract] [Full Text] [PDF]

				P. B. Bach Costs of Cancer Care: A View From the Centers for Medicare & Medicaid Services J. Clin. Oncol., January 10, 2007; 25(2): 187 - 190. [Abstract] [Full Text] [PDF]

				D. Schrag and M. Hanger Medical Oncologists' Views on Communicating With Patients About Chemotherapy Costs: A Pilot Survey J. Clin. Oncol., January 10, 2007; 25(2): 233 - 237. [Abstract] [Full Text] [PDF]

Comments: